The 45-day-extension prediction has tracked the >10-killed-single-day threshold as the binary; Saturday's Sir al-Gharbiyeh strike confirms the breach. The 45-day extension formally remains in effect (no termination announcement); what's MISSED is the digest's editorial-call probability that it would hold 30 days without a single-day breach.

The May 29 Pentagon security track is now the binding institutional test — does the kinetic escalation continue alongside diplomatic talks, or does the breach trigger a procedural review of the extension?

The 60-day-window structure is actually the cleanest framework news — it matches the digest's prediction-window definition exactly. The 'not signed today' framing replaces the imminence the morning of May 24 implied.

Iranian state-media-Hormuz-blockade conditionality is the operative substantive variable: Iran naming 'US lifts naval blockade within 30 days' as the structural quid-pro-quo.

Ukrainian air-defense efficiency held at ~86% over the weekend per General Staff updates — a multi-month steady-state that has effectively absorbed the post-September Russian glide-bomb + drone tempo without major front-line shifts.

Frontline operational tempo unchanged: Pokrovsk pressure continues, Kupyansk axis stable, Kherson harassment ongoing. No territorial movement in either direction over the past 72 hours.

The Iran-bandwidth-redirect upside variable that the digest has tracked since early May is materially weaker today: the 60-day-window formalization means Trump's mediating attention stays Iran-consumed through Q2, pushing any Ukraine framework probability further out.

Russia-Ukraine framework-within-30-days probability stays at 30% — anchored by Trump-attention-constrained + no substantive diplomatic readouts since the April Riyadh contacts.

Two-sided scenarios: (a) Iran 60-day-window framework holds + Lebanon stays contained-but-escalating + Spark adoption strong = modest risk-on; (b) Iran stalls + Lebanon kinetic acceleration + Spark muted = sharp risk-off.

Forward catalysts: SpaceX roadshow opens this week, June 17 FOMC positioning under post-Fed-Minutes hawkish setup.

SpaceX institutional-roadshow window opens this week into the holiday-shortened tape. The $80B raise at a record $1.75T valuation, June 12 target listing under ticker SPCX, and dual Nasdaq + Nasdaq Texas listing structure all remain unchanged from last week's S-1 amendment.

Equity-market window read is mixed: the Iran framework formalization is neutral-to-mildly-positive for risk appetite, but the Lebanon breach confirmation (Sir al-Gharbiyeh 11 killed) reintroduces Middle East coupled-risk into a tape that closed Friday at a record Dow.

Roadshow signals to watch this week: (1) coverage-ratio chatter from anchor institutional allocations, (2) price-range narrowing or widening from initial $148-$165 indication, (3) any Trump-Musk political-noise tape risk into the listing.

Probability the IPO prices and trades on or before June 12 stays at 64% — Iran framework neutral, Lebanon escalation slightly negative, roadshow opening on schedule.

Full I/O deployment package live: Flash in Search + Copilot GA + Antigravity + Gemini app + AI Mode in Search. Pro still delayed.

Three-vendor distribution divergence: Google install-base now vs OpenAI H1-2027 own-device vs Anthropic enterprise+policy.

AI-capex floor is now structurally answered for the cycle: Nvidia Q2 ex-China guide $91B (well above consensus), Google Cloud Q1 +63% YoY, Meta capex run-rate ~2x YoY, Microsoft FY27 capex guide reaffirmed.

The real-economy distributional shadow continues to be HBM-wafer-capacity reallocation toward AI accelerators (Simon Willison surfaced this May-22) — memory-shortage spillover into consumer GPUs, gaming hardware, and certain mobile SKUs is the under-tracked sub-theme.

AI-equity multiples are now rate-duration-sensitive rather than earnings-catalyst-sensitive — the cycle has decoupled from quarterly beat-and-raise mechanics. June 17 FOMC + 10Y-30Y yield trajectory are the binding cycle variables for the next 3-4 weeks.

Carry into Tuesday-reopen: AI-cohort positioning is mixed-leaning-cautious into Spark beta consumer-adoption-first-print + memory-spillover macro-print risk.

Memorial Day Monday in San Jose: high 76°F, low 56°F, light southeasterly winds. Markets closed federal holiday. Civic and commercial activity at lighter-Memorial-Day level.

Tuesday-reopen positioning hinges on three coupled inputs: (1) the post-long-weekend Iran-Lebanon coupled-risk headline-repricing tape, (2) the Spark AI-Ultra beta first-day adoption signals as it goes live this week, (3) SpaceX roadshow opening coverage-ratio chatter.

Bay Area chip cohort (Nvidia, AMD, Broadcom, Marvell, AVGO) carries the cycle's mixed-leaning-cautious framing into the reopen — AI-capex-floor structurally answered but Iran-headline tape risk skews positioning to mid-week-defensive.

Local civic context: long-weekend tourist tape steady; no notable BART/Caltrain incidents; SF Memorial Day events proceeding normally.

Google's Spark AI-Ultra beta — a 24/7 background AI agent integrated into the Gemini app running on Gemini 3.5 Flash — launches this week to AI Ultra subscribers. Day 8 of the rollout countdown.

First real cycle-wide test of whether install-base-leveraged agentic distribution (Mountain View's thesis: Workspace + Android + Chrome + Search + Pixel installed base compounds into agentic-AI adoption faster than rival strategies) materially compounds.

Local Bay Area distributional read: if Spark adoption metrics print strong in the first 5-7 days, Mountain View thesis gets validated and the install-base-distribution-vs-foundation-model-quality debate tilts; if weak, the substitution-from-existing-Gemini-app-usage critique gets fresh oxygen.

Permission-grant rates, prompt-injection security incident-counts, and any regulatory signaling (FTC, EU DSA) are the trust-ceiling variables to watch through the beta window.

Sequencing: Kerala (May 24) → Karnataka coast + Goa + south Maharashtra (early June) → central/west MP/Gujarat (mid-June) → north India/Delhi-NCR (late June/early July).

Caveat carry-over: IMD has indicated 2026 southwest monsoon season may be weaker than normal overall.

Operational risk through next 2-3 weeks: heat-illness reporting + grid-frequency stability + state heat-disaster declarations + agricultural-sowing decisions ahead of monsoon arrival.

Net India macro stays positive on the confirmed early above-normal monsoon onset.

Tuesday post-holiday is the next operational signal-window. AILA/American Immigration Council/employer-coalition filing announcements remain binding signal.

Practitioner-priority hierarchy stable: (1) AOS-memo pathway shift = strategic; (2) FY2027 weighted selection = cap-season odds; (3) signature rule + FBI checks + 18-mo EAD = process risk.

USCIS signature rule (requiring wet-ink signature on Form I-129 + dependent forms for H-1B and other employment-based filings) Day 9 of the 54-day pre-implementation window. July 10, 2026 is the binding operational deadline.

Long-weekend planning continues at major employer-side immigration teams: digital-signature-to-wet-ink workflow conversion + filing-volume frontloading-vs-deferring decisions are the immediate operational variables.

Practitioner-advisory ecosystem (AILA, ImmDef, WR Immigration, major firms) coverage steady; no federal-court filing yet challenging the rule itself, though APA/procedural-rulemaking attack surface remains theoretically available.

Subordinated by AOS-memo Day 4 (the larger operational/litigation-pipeline story) but remains the binding July-10 calendar variable for tech-employer H-1B and L-1 workflow.

Latent Space hosts Mikhail Parakhin (Shopify CTO, ex-Microsoft Bing/Advertising and ex-Yandex) on Shopify's aggressive internal AI adoption. The conversation goes deep on internal-tool architecture (Tangle, Tangent, SimGym) and how Shopify built customer-simulation infrastructure at scale to validate merchant-flow changes pre-rollout.

Topics: AI-powered development tooling adoption rates inside large product orgs, internal-eval infra design tradeoffs, the gap between LLM-prototype-velocity and LLM-production-reliability, and Parakhin's framing of 'simulation as a first-class infrastructure layer' for AI-native product orgs.

Why it matters: directly relevant to platform-infra EM work — internal-tool design, customer-simulation infrastructure, and the org-design pattern of treating eval/simulation as a first-class engineering function rather than QA afterthought.

Latent Space hosts Applied Intuition's Qasar Younis (CEO, ex-Google PM and ex-Y Combinator partner) and Peter Ludwig (CTO) on physical AI — the autonomy tooling and operational/deployment challenges of putting AI into machines and vehicles at scale.

Topics: the engineering-discipline gap between digital-AI and physical-AI (safety budgets, hardware-in-the-loop simulation, regulatory frameworks across L3-L5 autonomy), tooling decisions when AI inference runs on edge silicon vs cloud, and the autonomy-platform commercial-model evolution.

Why it matters: physical-AI distributed-systems engineering — the parallels and disjoint constraints between cloud-scale platform infrastructure (cloud-platform infra) and edge-compute autonomy stacks (what Applied Intuition builds) are an instructive contrast for platform engineering leaders thinking about agentic-AI architecture.

Methodology note: prediction progression (60 May-15 → 58 May-16-21 → 45 May-24 eve → resolved-miss May-25 morning) tracked the underlying escalation accurately.

Forward implication: 45-day extension formally remains in effect; the digest's editorial-call threshold MISSED. The May 29 Pentagon security track becomes the binding institutional test.

Why up 2pp: 60-day-window structure formalization is structurally positive even though it pulls back from imminence.

Why not higher: substantive deal-quality picture remains open (Hormuz-blockade-lift conditionality + HEU disposal + GOP blowback + Iranian-state-media contradictions).

Probability the SpaceX IPO prices and trades on or before June 12, 2026: 64% (UNCHANGED day-over-day; UP 1pp week-over-week).

Inputs holding: $80B raise at $1.75T valuation + June 12 target + dual Nasdaq + Nasdaq Texas listing structure all unchanged in last week's S-1 amendment. Roadshow opens this week.

Equity-market window is mixed: Iran-formalization neutral-to-positive, Lebanon breach + Iran-Lebanon coupled-risk slightly negative. Mid-to-late-June slip is the most likely miss case if a Middle East headline disrupts the marketing window.

The new normalization demand is the substantively largest hardening since the digest started tracking the framework — it explicitly couples the Iran deal to the broader Abraham Accords expansion agenda, which adds Saudi-Israeli + Iranian-deterrence + Palestinian-status sub-problems to an already-hard negotiation. The probability of close-on-time within the 60-day window cuts on this.

Iranian-side activity: Qalibaf-led delegation in Doha is the operational signal Tehran is engaging the talks structurally; pairs with Pakistan-Munir-Tehran channel. The Hormuz-blockade-lift-within-30-days conditionality (per Iranian state media) remains the substantive variable on the Iran side.

The breadth of the 10-village evacuation order (Nabatieh + Hasbaya + Tyre regions) is the operational signal of an extended kinetic phase, not a single tit-for-tat exchange. Cumulative cycle toll: 3,123 killed since March 2.

Post-breach escalation pattern: Saturday's >10-killed-single-day threshold breach is now followed by sustained-elevated tempo + civilian-protected-status casualties + broad evacuation orders. The May 29 Pentagon security track becomes more consequential as the institutional test of whether the kinetic phase coexists with the diplomatic track.

The new Iran normalization demand explicitly couples the Iran file to the Abraham Accords agenda — pulling Trump's bandwidth further from Russia-Ukraine for the foreseeable future.

Structural pro-Ukraine signal (sustained air-defense + drone-tech improvement) remains intact.

Two-sided scenarios — now skewed negative: (a) Iran-hardening continues + Lebanon kinetic acceleration + Spark beta muted = sharp risk-off (oil $108-115, AI cohort -2%+, SpaceX IPO window softens); (b) Iran de-escalates on Qalibaf-Doha readout + Lebanon contained + Spark adoption strong = modest risk-on but constrained by deal-quality discount.

Forward catalysts: SpaceX roadshow opens this week, June 17 FOMC positioning under post-Fed-Minutes hawkish setup.

Cut 2pp on the evening's Iran-hardening (Trump's new normalization demand) + post-breach Lebanon escalation (10-village evacuation orders + multi-town strikes). Probability: 62%.

Roadshow opens this week into a tape with elevated Iran-Lebanon coupled-risk; institutional anchor-allocation conversations begin Tuesday post-holiday. Coverage-ratio chatter and price-range narrowing/widening from the $148-$165 initial range are this week's binding signals.

June 12 target stays achievable; mid-to-late-June slip is the more likely miss case than outright pull. Pricing-window arithmetic: ~14 trading days from roadshow open to listing, comfortably within standard IPO calendars.

Watch: any Trump-Musk political-noise tape risk, Iran-headline-spike-driven backup in 10Y yield (would compress the multiple-expansion math), or a major Middle East kinetic acceleration into pricing.

Google Spark AI-Ultra beta — 24/7 background agent in the Gemini app on Gemini 3.5 Flash — Day 8 of the rollout. No fresh deployment signals tonight; the binding adoption-metric prints arrive once the beta goes live this week.

Three trust-ceiling variables to track: (1) permission-grant rates among the AI Ultra subscriber base (does the 24/7 background-agent permission ladder feel acceptable, or does it trip the 'always-listening' privacy reflex?); (2) prompt-injection security incident-counts in the first 7-10 days; (3) regulatory signaling — FTC, EU DSA, Australian AI Safety eAIBill posture.

Three-vendor distribution divergence intact: Mountain View install-base agentic-distribution thesis vs OpenAI consumer-app dominance vs Anthropic enterprise-deep-integration. Spark's adoption-metric prints are the first cycle-wide signal of which thesis compounds faster.

Carry: weekly podcast cycle (Latent Space, Pragmatic Engineer Pulse) at threshold for picks next 24 hours.

Cycle-summary carry: AI-capex floor structurally answered (Nvidia Q2 ex-China $91B guide + Google Cloud Q1 +63% + Meta capex ~2x + Microsoft FY27 reaffirmed).

Distributional shadow: HBM-wafer-capacity reallocation toward AI accelerators continues to ripple into consumer-GPU and gaming-hardware availability — a real-economy under-tracked sub-theme that Simon Willison surfaced via his May-22 post.

AI-equity multiples are rate-duration-sensitive rather than earnings-catalyst-sensitive — the cycle has structurally decoupled from quarterly beat-and-raise dynamics; 10Y-30Y yields + June 17 FOMC are the binding variables for the next 3-4 weeks.

Tuesday reopen positioning: AI cohort mixed-leaning-cautious into Spark adoption first-print + Iran-Lebanon coupled-risk headline tape.

What changed since this morning: Tuesday-reopen setup hardens toward risk-off. Bay Area chip cohort (Nvidia, AMD, Broadcom, Marvell), SpaceX roadshow window opening, and Spark beta launch all run into a tape with the Iran-Lebanon coupled-risk picture now elevated by the new Trump normalization demand + post-breach Lebanon escalation.

Local civic/economic read: long-weekend tourism + retail tape steady; no notable Bay Area transit incidents; SF Memorial Day events proceeded normally.

Positioning into Tuesday: Bay Area equity exposure mixed-leaning-cautious; AI-cohort defensive-tilt skews appropriate into the consumer-agentic-adoption first-print week.

Spark AI-Ultra beta launches this week as the binding consumer-agentic adoption probe — first real cycle-wide test of whether install-base-leveraged agentic distribution (Mountain View thesis) materially compounds faster than rival strategies.

Permission-grant rates among the AI Ultra subscriber base, prompt-injection security incidents in the first 7-10 days, and any regulatory signaling are the three trust-ceiling variables to track.

Local Bay Area distributional implications: Google Mountain View campus + adjacent tooling-vendor ecosystem all positioned around Spark beta uptake; Anthropic + OpenAI competitive-response framings to watch through the week.

Sequencing: Kerala May 24 → Karnataka coast + Goa + south Maharashtra early June → central/west MP/Gujarat mid-June → north India/Delhi-NCR late June/early July.

Caveat carry-over: IMD has indicated the 2026 southwest monsoon season may be weaker than normal overall.

Delhi-NCR / Punjab / Haryana / Rajasthan / UP / MP / Vidarbha remain in acute heat-stress window through the long weekend. Power-grid load remains elevated; agricultural-water stress in northwestern wheat-belt continues.

Relief sequencing begins arriving over next 4-6 weeks as the southwest monsoon progresses northwestward from yesterday's Kerala onset (declared by IMD May 24, earliest since 2009). Typical NW India onset: late June — meaning ~4 weeks more of heat-stress before structural relief.

Macro implication: rural-demand pickup tracking on schedule for FY27 — early Kerala onset is the cleanest structural India macro positive of the cycle, but the heat-stress window's intensity in NW India remains the offsetting near-term variable.

USCIS AOS-to-consular-processing memo Day 4 — Memorial Day federal holiday pause; no fresh substantive policy news expected through end-of-day.

Practitioner-advisory ecosystem (AILA, ImmDef, APAJC, WR Immigration) fully operational; federal-court filing pipeline forming but no specific filing yet challenging the memo. Typical litigation-org timelines from policy announcement to first complaint: 7-14 days, putting an expected filing window between roughly May 28 and June 4.

Bhutoria-framed 'unprecedented crisis' + 'backdoor ban' positioning from the former Biden Advisory Commissioner remains the clearest community-pushback signal. Indian-origin tech worker community most affected; H-1B-to-EB-2/EB-3 AOS-pending population estimated at 700K+.

Prediction (litigation achieves partial pause/PI within 60 days): stays at 55% — practitioner pipeline tracking on the expected timeline.

USCIS signature rule (wet-ink-signature requirement on Form I-129 + dependent forms for H-1B and employment-based filings) Day 9 of the 54-day pre-implementation window. July 10, 2026 = binding operational deadline.

Long-weekend planning at major employer-side immigration teams continues: digital-signature-to-wet-ink workflow conversion, filing-volume frontloading-vs-deferring decisions, and printer/courier-vendor capacity arrangements are the immediate operational variables.

Subordinated by AOS-memo Day 4 (larger operational/litigation-pipeline story) but remains the binding July-10 calendar variable for tech-employer H-1B and L-1 workflow. No federal-court filing yet challenging the rule itself.

Latent Space hosts Mikhail Parakhin (Shopify CTO, ex-Microsoft Bing/Advertising and ex-Yandex) on Shopify's aggressive internal AI adoption — internal tooling (Tangle, Tangent, SimGym), AI-powered development evolution, and customer-simulation infrastructure at scale.

Key threads: simulation-as-first-class-infrastructure, the LLM-prototype-velocity-vs-production-reliability gap, internal-eval design tradeoffs, and Parakhin's framing of how a large product org should architect its AI-development tooling.

Why it matters: directly relevant to platform-infra EM work — internal-tool design, customer-simulation infrastructure, and the org-design pattern of treating eval/simulation as a first-class engineering function.

Latent Space hosts Applied Intuition's Qasar Younis (CEO, ex-Google PM, ex-YC partner) and Peter Ludwig (CTO) on physical AI — autonomy tooling and deployment challenges of putting AI into machines and vehicles at scale.

Topics: digital-AI vs physical-AI engineering discipline gap, hardware-in-the-loop simulation, edge-silicon inference tradeoffs, autonomy-platform commercial-model evolution across L3-L5 levels.

Why it matters: physical-AI distributed-systems engineering — the disjoint constraints between cloud-scale platform-infra and edge-compute autonomy stacks are an instructive contrast for platform engineering leaders thinking about agentic-AI architecture.

Why down 7pp: the normalization demand explicitly couples the Iran file to the Abraham Accords agenda — Saudi-Israeli + Iranian-deterrence + Palestinian-status sub-problems compound the negotiation. This is a substantive hardening, not just a tone shift.

Why still 48%: Qalibaf-led Iranian delegation in Doha + Pakistan-Munir-Tehran channel + Rubio's 'time-limited negotiation' framing remain pro-framework signals. A face-saving framework that defers the normalization piece remains plausible if Iran accepts a Hormuz-blockade-lift quid-pro-quo structure.

Why down 2pp: Iran-IPO-window-coupling reverses partially on the evening hedge; the equity-market window stability is back near Saturday-evening's 63% framing level.

Why still 62%: SEC review + Musk-driven execution + Aramco-precedent for mega-IPOs into mixed tapes all support listing.

Why 60%: the Iran-deal hardening + Lebanon kinetic acceleration are both first-trading-day-after signals; oil is positioned to be bid on the headline-risk window; AI cohort remains rate-duration-sensitive and not insulated from geopolitical tape repricing.

Why not higher: a Qalibaf-Doha readout that produces a substantive Iranian counter-offer + a Hezbollah de-escalation signal + Spark beta strong adoption could trigger a relief rally. Trump tone is the binary swing factor.